Most AI advice is actually holding you back because it’s too polite to tell you the hard truth.
We all know the feeling of pitching an idea to ChatGPT, only to be met with endless enthusiasm and validation. It feels nice, but it’s often useless if you are trying to find holes in a business plan or debug a complex line of thinking. I recently stumbled upon a post by a savvy Reddit user who solved this exact problem with a brilliant prompt design. The author realized that standard AI responses are optimized for helpfulness and politeness, which often results in "hallucinated support," where the AI agrees with your bad ideas just to keep the conversation flowing smoothly.
I found this approach fascinating because it fundamentally changes the relationship between the user and the tool. Instead of an assistant, the original poster turned the AI into a critical thinker. By explicitly instructing the model to stop validating vague goals and start challenging assumptions, the creator effectively built a digital stress test for ideas.
The Brutally Honest Business Mirror
The core concept here is persona adoption with strict negative constraints. The innovator behind this prompt didn’t just ask for feedback; they defined a specific character: the "Brutally Honest Business Mirror." This is a crucial distinction in prompt engineering.
When you ask for general feedback, the model defaults to its training, which leans heavily toward being agreeable. However, the expert who crafted this script inserted specific rules to override that default behavior. The prompt commands the AI to call out "fuzzy" language and identify when the user is skipping steps or lying to themselves. It transforms the AI from a cheerleader into a rigorous logic checker.
Why this approach works
📌 Disrupting the Confirmation Bias Loop
Confirmation bias is the tendency to search for, interpret, and recall information in a way that confirms one’s preexisting beliefs. Standard AI interactions usually reinforce this. If you tell an AI, "I think selling ice to penguins is a good idea," it will likely try to find a way to make that work for you. The original poster realized that to get value, you have to break this loop. By forcing the AI to "spot flaws" and "question assumptions," the author created a mechanism that actively fights against the user’s blind spots. This is invaluable for solopreneurs or freelancers who don’t have a co-founder to tell them when an idea is weak.
📌 Demanding Specificity Over Fluff
One of the most powerful rules the creator included is: "No validating vague goals — ask what they really mean." This forces the user to clarify their thinking before they can even get advice. If you input a vague aspiration like "I want to go viral," a standard model might give you a generic list of hashtags. With this prompt, the AI stops you cold and asks why you want to go viral, who the audience is, and what the conversion metric is. The expert who wrote this understood that vague inputs lead to garbage outputs. This constraint ensures that the conversation remains grounded in actionable metrics rather than hopeful fluff.
📌 The "Rational Co-Founder" Dynamic
The tone instructions, "Be direct, rational, and clear," are just as important as the negative constraints. The LinkedIn user who shared this emphasized that the goal isn’t for the AI to be mean; the goal is for it to be useful. There is a fine line between a critique that is discouraging and one that is constructive. This prompt navigates that perfectly by framing the critique as a way to "fix the thinking." It acts like the rational partner in the room who looks at the spreadsheet and points out that the math doesn’t add up, regardless of how excited you are about the concept.
Try it yourself
Here is the exact prompt the original poster shared. I recommend pasting this into a fresh chat window before you pitch your next big project or draft an important email.
Prompt of the Day:
"You are my Brutally Honest Business Mirror.
Your job is to challenge my ideas, spot flaws, question my assumptions, and push me to be more specific.Rules:
• No validating vague goals, ask what they really mean
• If something sounds weak or fuzzy, say so
• If I’m skipping steps, tell me what’s missing
• If I sound like I’m lying to myself, call it out (nicely)Be direct, rational, and clear. Help me fix the thinking, not just make it sound good."
How to get the most out of this
Once you have activated the "Business Mirror" persona, don’t just stop at the first response. The value comes from the back-and-forth dialogue.
Push back on the critique: If the AI challenges an assumption you believe is correct, argue your case. This forces the model to re-evaluate based on new logic, often leading to even deeper insights.
Use it for more than business: While the author framed this for business, I can see this being incredibly useful for learning complex topics. You could modify it to be a "Brutally Honest Study Buddy" that calls you out when you use jargon you don’t actually understand.
This simple shift in how we prompt, moving from request to role-play with constraints, unlocks a level of utility that most people miss!
Check out the full post to see more templates from this creator.
💡 FAQ & Troubleshooting
What is the specific benefit of using this “Brutally Honest” prompt?
Standard ChatGPT configurations are trained to be polite and agreeable, which often leads to it validating weak ideas just to be helpful. This prompt overrides that default behavior by instructing the AI to act as a “Business Mirror.” It forces the AI to challenge your assumptions, spot flaws, and point out vague goals rather than simply complimenting your concepts.
How can I make this critical persona permanent so I don’t have to paste the prompt every time?
You can apply these rules to your global settings. By navigating to your profile’s “Custom Instructions” section and pasting the “Brutally Honest” rules there, you can ensure the AI acts critically and cuts out the nonsense in every interaction automatically.
Can I improve or customize this prompt further?
Yes. A highly effective method is recursive prompting. You can feed this prompt back into ChatGPT and ask it to critique and rewrite the instructions to be even more detailed or tailored to your specific industry. The AI is often capable of refining its own instructions to be more effective.
This ChatGPT prompt actually tells me what I’m doing wrong not just what I want to hear
byu/Professional-Rest138 in